The purpose of this study is to test the effect of financial performance on earnings management, with the presence of a firm’s specific characteristics, in a non-credible financial information context such as Tunisia. A panel data approach, namely multiple regression analysis, was applied on a sample of 30 firms operating in different sectors and observed over the period 2012–2021. Two estimation methods, the fixed effects and random effects models, are used. To choose the best estimation method, the Breusch-Pagan and Hausman tests were used.The results indicate, on the one hand, the financial performance measured by Tobin’s Q and Marris’ ratio, positively affects earnings management. On the other hand, the variables of firm characteristics, such as financial leverage, asset structure, growth rate and sectoral affiliation, decrease earnings manipulation, and firm size and managers’ ownership have no effect on earnings management. This means that managers of Tunisian firms manipulate their results to improve the level of performance and their financial sate. This manipulation is driven by goals other than those observed in other contexts and related to the financial market. This finding contributes to the literature on the association between several features of earnings management and firm performance, with the effect of firm characteristics.